scook
R-Mix sounds perfect for this application
I know R-Mix didn't help in this case because it doesn't have the ability to zoom in tight enough to isolate the frequencies that Robert needed.
However, I wanted to put in a plug for R-Mix. I'm working on a big band arrangement for a signer. I have been trying to get a voice track from him so I will have a good idea how he phrases things -- I can keep the horn punches out of his way. We were never able to connect in a normal sort of way, so my fallback plan was to catch him at a rehearsal and have him sing into a field recorder, with an MP3 player in his ears.
It seemed like a good plan until the guy with the key to the union hall never made it, so the band was standing outside waiting for 45 minutes, in a noisy environment. I figured I might as well try to capture that recording. There was a a lot of background noise, especially other band members talking right next to him, and the noise of traffic driving by.
But R-Mix was able to really isolate the vocal pretty well. By adding a noise gate to the chain, It actually was a pretty clean recording -- a lot better than I expected.
The only problem is that because of the noise, he wasn't in sync with the music all the way, but I can tell what he was trying to do I think. And then he said my MP3 track was about 20 beats too slow. So my next move is to do the 30-day trial of Melodyne. Looking at the videos, I ought to be able to put his vocals right where (I think) he intended them, and do it at the faster tempo.
OK, it would be a lot faster to drive over to his house and just record the *^%$#% thing again, but this is war. I'm determined to make my plan work. :) But seriously, if Melodyne can really do what it says, this would be an awesome tool, so I am eager to try it out on this low risk project.
And speaking of noise removal, if Melodyne can actually find all the real notes, then it seems like there ought to be a way to throw away everything that Melodyne didn't consider to be real notes.